Gemini CLI

Gemini CLI supports remote MCP servers over Streamable HTTP natively, so no proxy or npx command is needed.

Prerequisites

Gemini CLI installed
A MCPCore server with at least one tool

Quick install

On the server’s Integration tab, select Gemini CLI and copy the command from Quick install:

Public
Unlisted
API Key

gemini mcp add --transport http my-server https://my-server.mcpcore.io/mcp

gemini mcp add --transport http my-server https://my-server.mcpcore.io/mcp/<secret-key>

export MCPCORE_API_KEY="<your-api-key>"
gemini mcp add --transport http my-server https://my-server.mcpcore.io/mcp --header "Authorization: Bearer $MCPCORE_API_KEY"

The command reads the key from the MCPCORE_API_KEY variable of your shell. Turn on Include my API key to get the command with the key filled in.

Manual setup

Merge the server into ~/.gemini/settings.json (or .gemini/settings.json in your project). The Manual setup snippet on the Integration tab looks like this:

{
  "mcpServers": {
    "my-server": {
      "httpUrl": "https://my-server.mcpcore.io/mcp"
    }
  }
}

For API Key mode:

{
  "mcpServers": {
    "my-server": {
      "httpUrl": "https://my-server.mcpcore.io/mcp",
      "headers": {
        "Authorization": "Bearer <your-api-key>"
      }
    }
  }
}

Verify

Start Gemini CLI and run /mcp. Your server and its tools should be listed. Ask Gemini to use one of your tools to confirm it works.

Troubleshooting

Problem	Solution
Server is disconnected	Check the URL and that the server is active
`401 Unauthorized`	Check the `Authorization` header and that `MCPCORE_API_KEY` was set when you ran the command
JSON parse error	Validate the syntax of `settings.json`
Tool calls fail	Check Traffic Logs

Getting Started

MCP Servers

Tools

Observability

Connect AI Clients

Prerequisites

Quick install

Manual setup

Verify

Troubleshooting

​Prerequisites

​Quick install

​Manual setup

​Verify

​Troubleshooting

Prerequisites

Quick install

Manual setup

Verify

Troubleshooting